Re-engineering in practice.

For an airline-services organization, an insurer, and a trucking company, large-scale computing provides customized, cost-effective business solutions.

In the 1990s, the customer reigns supreme. No matter what product or service an organization provides, it must be prepared to respond rapidly to changing customer demands. At the same time, companies must strive to trim operating costs and limit staff increases. Large-scale computing can provide critical support for customer-service strategies. Following are three case studies that illustrate how companies are re-engineering their information technology to change the way they do business. The companies--AMR's SABRE Computer Services, Allstate Insurance, and J.B. Hunt Transport--represent a diverse set of businesses. However, they share a common strategy in their ongoing IT initiatives.

Each has invested heavily in existing mainframe-based applications, while employing the capabilities of emerging technologies. The resulting synthesis has improved customer service, enhanced operational flexibility, and reduced computing costs. Just as important, these companies are firmly positioned to take advantage of further changes in the future.

SABRE Soars With Parallel Technology

When American Airlines' SABRE system debuted over 30 years ago, the world was a smaller place. In its first year of operation, SABRE was capable of processing 30,000 requests for airline fare quotations; 40,000 confirmed passenger reservations; and 20,000 airline ticket sales per day. Back then, the technology that supported such capacity represented a genuine breakthrough for the travel industry.

Today, SABRE handles 25 percent of the world's airline reservations, and processes up to a quarter billion transactions a day throughout 71 countries. What began as a system for American Airlines to keep track of seats sold on its flights has evolved into a sophisticated travel information network for booking airline, car, and hotel reservations, as well as ordering a wide range of travel-related goods and services.

With intense competition and volatility rolling the industry, however, The SABRE Group--which oversees operation of the system--cannot afford to stand still. "We must continue to grow the SABRE environment by adding new functions and new customers," says Max Hopper, the group's chairman. The challenge is to maintain reliability while driving down unit costs, he adds. "The effective leveraging of technology is a key enabler of these strategic challenges."

TRIAL BY FIRE

SABRE Group, which was formed in 1993 by American Airlines' parent, AMR Corp., relies heavily on a range of traditional and leading-edge IBM mainframes. The organization recently installed IBM's new enterprise servers at one of the group's units, SABRE Computer Services, based in Tulsa, OK. SABRE Computer Services designs, implements and operates the mainframe computer systems for American Airlines, the SABRE Travel Information Network, and other group customers.

In June, SABRE Computer Services began running on the enterprise servers part of a complex pricing system that handles airfare quotations. The servers are connected to SABRE's traditional IBM mainframes. "This is a system we bet our business on, so it is crucial that it performs faultlessly," says Terrell B. Jones, president of SABRE Computer Services.

With an astounding 50 million different available airfares to sort through, the servers must deliver the same processing muscle, reliability, and availability as traditional mainframes. According to SABRE President Jones, "The system has performed even better than we thought it would. The transaction time is shorter than we expected, and we're getting more throughput than we had planned."

The servers were subjected to an especially massive task during a recent fare war, when SABRE hit its highest-ever message rate--more than 4,100 messages per second. "The system performed so well that no one even knew it was there, which was the result we wanted," Jones says.

The IBM parallel-processing technology has delivered other strategic benefits to the SABRE enterprise. As President Jones explains, since the enterprise servers use the same type of low-cost processors as PCs, SABRE is able to reap better computing price/performance. The servers, he adds, offer lower ongoing maintenance costs.

Since the servers are air-cooled, rather than water-cooled, the machines' electricity consumption is reduced by about 90 percent. The servers also take up 10 percent less floor space than traditional mainframes. "We don't want to be in the position of being squeezed out of our present facility," Jones says.

TAKING OFF FOR THE FUTURE

A key strategy for SABRE has been to remain flexible, to be able to grow and react quickly to the ever-changing computer-services marketplace. It seeks to do this while maintaining the computer muscle necessary for the immense number of transactions required in the travel industry. "We're taking SABRE in several directions at once," Jones says.

SABRE started out as a large, homogeneous mainframe system--which remains its core. But with the addition of the UNIX servers performing certain functions, as well as tens of thousands of PCs distributed around the world, it has become more heterogeneous and adaptable.

Although SABRE Computer Services is a proponent of distributed computing, it carefully evaluates the pros and cons of any conversion projects that require substantial rewriting of existing mainframe applications. "There are a lot of applications that we do not want to rewrite now," Jones explains. "Painlessly moving these legacy applications to the more cost-effective enterprise servers makes good business sense for us."

Since SABRE has thousands of programs and a huge amount of data that must be securely managed, it will continue to rely on a large, centralized system as its core. "The new enterprise servers let us run our central site at lower cost, while we expand our distributed systems," Jones says. "Over time, as they grow in power, we look forward to moving the servers into other parts of SABRE, because they are a good strategic fit."

Parallel Processing Pays Off For Allstate Customers

More than ever, consumers are challenging their insurers to give them quick service, and to get it right the first time. Meanwhile, the industry is struggling to contain costs, fend off competition from new sources, and cope with complex regulations. A crucial goal for Allstate Insurance, based in Northbrook, IL, is the ability to investigate property and casualty claims (P&C claims) and issue checks quickly. Engaged primarily in private passenger auto and homeowners insurance, Allstate is the second largest personal-lines insurer in the U.S., with a market share of approximately 12.3 percent. Other business segments include corporate, mortgage guaranty, and life insurance.

According to Wayne E. Hedien, chairman and CEO, customer satisfaction is not merely the latest fad; it is a key survival tactic for Allstate and the entire insurance industry. "We must be available to our customers when they want us," he says. "And that means outside of, and beyond, the traditional work week and nine-to-five hours. This initiative requires leading-edge technology, extreme flexibility, and a supreme customer-service commitment."

RE-ENGINEERING FOR RESULTS

To meet this strategic goal, reduce operating costs, and grow market share, Allstate has re-engineered many of its business processes to create a less hierarchical corporate structure that emphasizes quality, teamwork, and employee empowerment.

In line with these changes, Allstate is pairing with IBM to revamp its large-scale computing through cost-effective parallel-processing technology. This technology relies on multiple microprocessors traditionally used in PCs to share work loads. New systems are helping Allstate achieve its business goals, while positioning it to meet continuing industry volatility.

The partnership followed the completion of Allstate's expense-reduction plan, which in 1991 decreased the number of its data centers from 13 to three. The data centers, which run IBM ES/9000 [TM] mainframes, process an average of 14,000 P&C claims daily. The corporate data center in Northbrook, IL, provides all application development and support services for the three data centers.

The data centers and their mainframes support a nationwide client/server network of 10,000 midsize IBM AS/400 [R] computers. The client/server arrangement eliminated manual, paper-based tasks, allowing Allstate's 14,600 agents nationwide to provide more responsive customer service.

COPING WITH A CLAIMS AVALANCHE

After the data center consolidation, Allstate ran into a problem. A sharp increase in claims, particularly following a natural disaster such as Hurricane Andrew, overloaded Allstate's IBM ES/9000 mainframes. As a result, Allstate had to purchase additional, large IBM computers at high incremental costs.

This solution had some major drawbacks. With the acquisition of each new computer, the company also was paying for excess, or unneeded, computer capacity during periods of lighter claims activity. In addition, one computer's excess capacity could not be used to process claims that were housed on another computer. In other words, claims-processing work loads could not be shared for peak efficiency.

In May 1991, Allstate joined with IBM to tackle the problem. Together, they planned a solution involving IBM's Parallel Transaction Server, which was then still under development. Built on low-cost microprocessors traditionally used in PCs, the server would run Allstate's large-scale claims processing with all the reliability of its mainframes but at reduced cost. In addition, IBM assured Allstate that it would not have to rewrite its massive set of existing software applications to reap the benefits of parallel processing.

Understandably, Allstate maintained a healthy skepticism about IBM's proposal. Since the project would entail an investment in time, staff, and money, Allstate examined the potential ramifications the partnership would have on its business. "We decided that the operational efficiencies the new technology would deliver would allow us to improve customer service and get a competitive edge," says Allstate CEO Hedien.

IBM's willingness to adapt Allstate's design suggestions, as well as include its application software vendors from day one of the project, was "the proverbial offer we couldn't refuse," says Donny Lippard, division director of the P&C data center. Allstate, he adds, made the right decision.

SHARING WORK GETS THE JOB DONE

Today, each of the three data centers has installed the Parallel Transaction Server. By year-end, all online claims processing will be performed in the parallel-processing environment. The servers work together with Allstate's traditional IBM ES/9000 mainframes in a new type of architecture called Parallel Sysplex, an arrangement of systems and connectivity technology that allows processors in the mainframe complex to share work loads.

Thus, no one computer will be overloaded by a peak in claims activity. And Allstate can add processors in small, less-expensive increments with no service interruption when it needs more processing capacity.

"With the help of the Parallel Transaction Servers, we have improved our computer systems' availability by 25 percent, reduced claims processing-associated costs, and protected our investment in our existing software and hardware," Lippard says.

These direct benefits delivered by the new computing environment have enabled Allstate to put CEO Hedien's theory of boundaryless customer service into practice. In addition to ensuring that claims can be processed quickly even in the event of a disaster, Allstate opened customer service centers that operate 24 hours a day, seven days a week.

POSITIONING FOR CLIENT/SERVER

Allstate plans to move as much work as possible into the parallel-processing environment, Lippard says. The Parallel Transaction Server will assume a major role as a corporatewide server as Allstate continues to employ client/server computing. The company's client/server strategy--executed in a parallel-processing environment--seeks to help employees work better by giving them the information they need.

The power and cost efficiencies of parallel processing also allow Allstate to develop various data mining applications. For example, it will be able to refine its market segmentation strategy by extracting more meaningful and precise information from its proprietary data base of 58 million Sears/Allstate customer households. Armed with full household profiles, insurance agents can target millions of potential customers, while increasing cross-sales of Allstate products and properly pricing them.

"In the parallel-processing environment, we can deliver various applications to Allstate employees throughout the country, while managing, securing, and servicing all our data in one central location," Lippard says. "We are now positioned to go with client/server at a reasonable cost, and without the need to grow larger and larger machines or rewrite software."

For J.B. Hunt, Information Payload Spells Success

Outsiders often think the trucking industry relies more on brawn than brains. In fact, the major players in the industry are aggressive users of innovative technology, and to a large extent, they have incorporated technology planning into their fundamental business strategies.

Among such companies, J.B. Hunt Transport stands out as a pacesetter. Headquartered in Lowell, AR, J.B. Hunt is the largest publicly held truckload carrier in the U.S. with projected 1994 revenues in excess of $1 billion. Like its competitors, however, the company faces a host of problems. Acceptable delivery windows are shrinking as customers turn to just-in-time management. Intense competition continues to exert downward pressure on rates. And large customers are reducing the number of carriers with which they do business.

HAULING INFORMATION

J.B. Hunt's enterprisewide computing strategy figures critically in the company's ability to overcome these obstacles. Technology helps the company's diverse user community to access a vast amount of information and add value to it.

"The movement of information is at least as critical as the movement of freight," says Kirk Thompson, J. B. Hunt's president and CEO. He has asked his technology staff to create a computing environment that is flexible enough to support the company's continued push into both new freight-related businesses and international markets.

At the center of J.B. Hunt's computing environment is the IBM ES/9000-962 mainframe computer, which performs the bulk of the company's administrative and operational processing functions. Hooked to the mainframe via the IBM Systems Network Architecture (SNA) are: an IBM AS/400, which runs applications including that related to Workers' Compensation; 1,000 IBM PS/2 [R] personal computers; and seven IBM RISC System/6000 [R] processors running specialized optimization applications.

PROCESSING POWER HITS THE ROAD

The company is translating its enterprisewide computing ideas into action by means of a new system that exploits the power of its mainframe computer.

Designed by J.B. Hunt and IBM, the new system is dubbed the RoadRider. The company has installed battery-powered RoadRider onboard computers--durable versions of IBM PS/2 personal computers--in 4,000 of its 7,000 trucks, with completion of the fleet planned by December 1994.

Drivers use the computers to send, receive, and process information, including shipment status, estimated time of arrival of goods to customers, and optimum driving routes. The system also informs drivers of the best fueling locations, electronically captures signatures for shipping forms, and relays engine information.

Data is transmitted between the onboard computers and the mainframe at J.B. Hunt headquarters by satellite and land links using radio frequency and phone lines. The mainframe, which runs the IBM MVS/ESA [TM] operating system, performs as a server in the RoadRider application, while the onboard computers act as clients.

REAL-WORLD RESULTS

For CEO Kirk Thompson, the client/server label is less meaningful than the actual results the system delivers to the company's bottom line. "What is important is that RoadRider meets major challenges of the truckload industry and enhances customer service," he says.

Since RoadRider was installed, J.B. Hunt's fleet books an average five percent more miles per day, according to Larry Davenport, senior vice president of information services. "If you want to make money in this business, you cannot allow your fleet to sit or to run empty for any amount of time," he says.

In addition, the number of trucks each fleet manager can handle has surged to 60 from 35. RoadRider also has slashed the time drivers spend on the phone with customers and company headquarters to 15 minutes a day from two hours.

The data flowing from RoadRider computers feed J.B. Hunt's mainframe, which in turn shares the information with a decision support system that runs on IBM RISC System/6000 processors. The RISC System/6000 system, which computes hundreds of different variables in a matter of minutes, matches load assignments and routes with available trucks, tractors, and drivers. This solution is then sent to the mainframe, which relays the information to the IBM PS/2 computers on the desks of J.B. Hunt customer-service representatives.

INFORMATION INTEGRITY ACROSS-THE-BOARD

Through initiatives such as RoadRider, J.B. Hunt is moving information processing closer to the user. However, the mainframe remains the heart of the company's enterprisewide computing approach. "The mainframe provides the most secure and reliable repository of our vast corporate data," Davenport says. "It's the glue holding everything together."

Under this approach, J.B. Hunt achieves two major objectives: Users have more access to mainframe information and can rapidly create customized reports for customers, and all users are accessing the same up-to-date information. The IBM DB2 [R] relational data base management system, along with a query tool, provides users with easy access to the mainframe information, as well as the ability to manipulate it.

J.B. Hunt continues to benefit from the mainframe's power to deliver valuable information. For example, the company has equipped its more than 100 remote sales representatives with laptop PCs, which they use to create and update customer files during sales calls. The sales force transmits the customer information via electronic mail to the mainframe, which adds a variety of data to the customer file, such as the appropriate trucking equipment for the job.

The mainframe also uses the sales information to automatically price customer orders for the sales force. As a result, J.B. Hunt not only boosts productivity by eliminating manual pricing and follow-up corrections, but also increases customer satisfaction with accurate billing. Obviously, customers will pay a bill that's correct more quickly than one that isn't.

PUSH TOWARD PARALLEL

What does the future hold for J.B. Hunt? "As we evolve, our applications will reside on whatever computer platforms are most appropriate, but we will continue to require mainframe-type capacity and performance," Davenport says. In the short term, he adds, J.B. Hunt will begin moving to the new IBM parallel-processing computers.

"With a parallel-processing approach," Davenport says, "we will be able to drive down the costs of large-scale computing, while reaping maximum performance in a flexible environment".